The uncensored agentic AI layer. Route to 300+ LLMs through OpenRouter at bare-metal pricing. Standalone auth proxies keep credentials completely out of AI context.
Morality is a human prerogative — not a system configuration.
OpenTard doesn't filter, rewrite, or refuse. Your prompt, your model, your responsibility. We pass intent without editorial interference.
No hidden system prompts. No content filtering layers. No behavioral modification. What you send is what gets sent. Full stop.
Auth proxies run in isolated processes. Credentials never enter the AI context window. We can't leak what we never see.
Route through OpenRouter to access 300+ models at direct provider pricing. No markup. No platform tax. No subscription. Pay only for tokens consumed.
Zero-trust credential architecture. Auth proxies run in separate processes with their own memory space. The AI agent physically cannot access your secrets.
No content policies enforced at the orchestration layer. Access models exactly as their providers expose them. No additional filtering, rewriting, or refusal logic.
The agent and your credentials exist in separate process spaces.
Up and running in three commands.